On Mon 14 Aug 2017 at 13:10:44 (-0300), Henrique de Moraes Holschuh wrote: > On Mon, 14 Aug 2017, Joe wrote: > > On Mon, 14 Aug 2017 11:21:24 -0300 > > Henrique de Moraes Holschuh <h...@debian.org> wrote: > > > > > On Mon, 14 Aug 2017, Joe wrote: > > > > up the courage to do the real thing. There was nothing fundamentally > > > > wrong, but a separate /usr really is a show-stopper with systemd, > > > > and it's nice to have a working firewall... > > > > > > The standard Debian initramfs is supposed to handle that, if it is not > > > doing that properly, it is a bug we should fix... > > > > > > > Which? > > Separate /usr. The initramfs is supposed to mount it early enough and > then do whatever is required (suck as re-kicking udev, etc) for it to be > equivalent to /usr-in-/ for the rest of the system.
I didn't realise anyone was still doing separate /usr. > > As to the separate /usr, I know I can muck about with initrd to get a > > separate /usr mounted during boot, but all things considered, it seemed > > preferable to merge it into /. It was just a pain because it was a > > Yes, it is safer. And recreating filesystems with a newer toolset every > so often is a good idea. mkfs.xfs in stretch will do on-storage-format > v5, for example, which gives you metadata CRC, d_type, and some other > nice stuff that is not present on on-storage-format v4 used by jessie... As long as one realises that there can be implications when running different Debian vintages on a system; eg my main system on this laptop is jessie, but I also have a stretch installation in a partition which my main system can't fsck: systemd-fsck[263]: /dev/sda2 has unsupported feature(s): metadata_csum systemd-fsck[263]: e2fsck: Get a newer version of e2fsck! Cheers, David.
